1 |
On 2020-06-22 18:18, Michał Górny wrote: |
2 |
> Hello, |
3 |
> |
4 |
> Here's another question for the Council nominees. I'd like to ask |
5 |
> the Council members to disclose their corporate affiliations, |
6 |
> in particular whether they are employed or in partnership with companies |
7 |
> using Gentoo or Gentoo derivatives. |
8 |
> |
9 |
> This is because I believe that the electorate deserves to know whether |
10 |
> their elected Council member may end up being in conflict of interest |
11 |
> between doing what's right by the wide community and what's requested by |
12 |
> his employer. |
13 |
> |
14 |
|
15 |
As you may or may not know I've been working at Adjust GmbH |
16 |
(www.adjust.com) for the last almost 5 years. |
17 |
|
18 |
Most of my job is babysitting a fleet of around 1000 servers running |
19 |
Gentoo, and ensure things work. I've had a few silly job titles, but I'm |
20 |
basically still Fixer of Things. (We're hiring!) |
21 |
|
22 |
This inherently makes me motivated to have things in a sane state - e.g. |
23 |
packages actually compiling, updates not breaking and other very |
24 |
outdated traditional ideas about software development. |
25 |
|
26 |
(And this is why I'm against things like the current py2 purge: There is |
27 |
code out there that works, can't be rewritten to py3 in a reasonable |
28 |
time*, and hasn't been rewritten in another language yet. There is no |
29 |
fundamental reason to exorcise all things older than 6 weeks, and it |
30 |
just forces me to spend time on useless busywork instead of doing |
31 |
something useful. |
32 |
And it's inconsistent - packages like chromium won't get masked, because |
33 |
... err... ? |
34 |
But I don't have the time to fight against this madness, so I just move |
35 |
everything useful to an overlay where it is vandalism-safe. Somehow that |
36 |
doesn't sound like a smart strategy to me but what can you do) |
37 |
|
38 |
This doesn't mean I'm a statist, progress can be nice, but these days |
39 |
it's both computationally expensive with some packages taking more than |
40 |
a cpu-day to build, and lots of breakage because very few upstreams to |
41 |
anything resembling software engineering. |
42 |
(e.g. gcc breaking ABI (wtf gcc5), glibc breaking collation (glibc 2.28 |
43 |
which makes updating things exquisitely super fun times), random |
44 |
packages bundling in LLVM, openssl** and whatever else looks cute) |
45 |
|
46 |
tl;dr: I want to be lazy, so stop breaking stuff ;) |
47 |
|
48 |
Have fun, |
49 |
|
50 |
Patrick |
51 |
|
52 |
|
53 |
|
54 |
* "can't be rewritten" - there's some corners of python like |
55 |
manipulating binary data that are not cleanly portable to py3, and the |
56 |
people who would do the rewrite-from-scratch prefer using other |
57 |
languages like Go that don't mutate as fast (since rewriting sucks); as |
58 |
such this legacy py2 code will exist until it is either no longer |
59 |
needed, or the cost of rewriting is smaller than the negligible cost of |
60 |
maintenance. |
61 |
|
62 |
** yes, a bundled-in security issue. Isn't it great! |